feat(style): pushing uncommited changes

Australian-Structural-Biology-Computing · Oct 10, 2024 · 9af55b1 · 9af55b1
1 parent 5fe8e1d
commit 9af55b1
Show file tree

Hide file tree

Showing 3 changed files with 57 additions and 45 deletions.
diff --git a/modules/local/run_alphafold2.nf b/modules/local/run_alphafold2.nf
@@ -5,9 +5,12 @@ process RUN_ALPHAFOLD2 {
     tag "$meta.id"
     label 'process_medium'
 
-    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'docker://nfcore/proteinfold_alphafold2_standard:1.0.0' :
-        'nfcore/proteinfold_alphafold2_standard:1.0.0' }"
+    // Exit if running this module with -profile conda / -profile mamba
+    if (workflow.profile.tokenize(',').intersect(['conda', 'mamba']).size() >= 1) {
+        error("Local RUN_ALPHAFOLD2 module does not support Conda. Please use Docker / Singularity / Podman instead.")
+    }
+
+    container "nf-core/proteinfold_alphafold2_standard:1.1.1"
 
     input:
     tuple val(meta), path(fasta)
@@ -19,7 +22,7 @@ process RUN_ALPHAFOLD2 {
     path ('mgnify/*')
     path ('pdb70/*')
     path ('pdb_mmcif/*')
-    path ('uniclust30/*')
+    path ('uniref30/*')
     path ('uniref90/*')
     path ('pdb_seqres/*')
     path ('uniprot/*')
@@ -34,31 +37,35 @@ process RUN_ALPHAFOLD2 {
 
     script:
     def args = task.ext.args ?: ''
-    def db_preset = db_preset ? "full_dbs --bfd_database_path=${params.bfd_dir_path}${params.bfd_metaclust_clu_complete_id30_c90_final_seq_sorted_opt_name} --uniclust30_database_path=${params.uniclust30_dir_path}${params.uniclust30_db_name}" :
-        "reduced_dbs --small_bfd_database_path=${params.small_bfd_path}${params.bfd_first_non_consensus_sequences_name}"
+    def db_preset = db_preset ? "full_dbs --bfd_database_path=${params.alphafold2_db}/bfd/bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt --uniref30_database_path=${params.alphafold2_db}/uniref30/UniRef30_2021_03" :
+        "reduced_dbs --small_bfd_database_path=${params.alphafold2_db}/small_bfd/bfd-first_non_consensus_sequences.fasta"
     if (alphafold2_model_preset == 'multimer') {
-        alphafold2_model_preset += " --pdb_seqres_database_path=${params.pdb_seqres_dir_path}${params.pdb_seqres_txt_name} --uniprot_database_path=${params.uniprot_dir_path}${params.uniprot_fasta_name} "
+        alphafold2_model_preset += " --pdb_seqres_database_path=${params.alphafold2_db}/pdb_seqres/pdb_seqres.txt --uniprot_database_path=${params.alphafold2_db}/uniprot/uniprot.fasta "
     }
     else {
-        alphafold2_model_preset += " --pdb70_database_path=${params.pdb_dir_path}${params.pdb70_name} "
+        alphafold2_model_preset += " --pdb70_database_path=${params.alphafold2_db}/pdb70/pdb70_from_mmcif_200916/pdb70 "
     }
     """
-    if [ -f ${params.pdb_seqres_dir_path}/${params.pdb_seqres_txt_name} ]
-        \$PDB_SEQRES_TEMP=\$(mktemp --directory)
-        cp ${params.pdb_seqres_dir_path}${params.pdb_seqres_txt_name} \${PDB_SEQRES_TEMP}/
-        then sed -i "/^\\w*0/d" \$PDB_SEQERS_TEMP/${params.pdb_seqres_txt_name}
+    RUNTIME_TMP=\$(mktemp -d)
+    nvcc --version 2>&1 | tee /home/z3545907/nvcc.txt
+    nvidia-smi 2>&1 | tee /home/z3545907/nvidia-smi.txt
+    if [ -f ${params.alphafold2_db}/pdb_seqres/pdb_seqres.txt ]
+        cp ${params.alphafold2_db}/pdb_seqres/pdb_seqres.txt \${RUNTIME_TMP}
+        then sed -i "/^\\w*0/d" \${RUNTIME_TMP}/pdb_seqres.txt
     fi
-    if [ -d ${params.alphafold2_params_path} ]; then ln -r -s ${params.alphafold2_params_path} params/; fi
+    if [ -d ${params.alphafold2_db}/params/ ]; then ln -r -s ${params.alphafold2_db}/params params; fi
     python3 /app/alphafold/run_alphafold.py \
         --fasta_paths=${fasta} \
         --model_preset=${alphafold2_model_preset} \
         --db_preset=${db_preset} \
         --output_dir=\$PWD \
         --data_dir=\$PWD \
-        --uniref90_database_path=${params.uniref90_dir_path}uniref90.fasta \
-        --template_mmcif_dir=${params.template_mmcif_dir} \
-        --obsolete_pdbs_path=${params.obsolete_pdbs_path} \
+        --uniref90_database_path=${params.alphafold2_db}/uniref90/uniref90.fasta \
+        --mgnify_database_path=${params.alphafold2_db}/mgnify/mgy_clusters_2022_05.fa \
+        --template_mmcif_dir=${params.alphafold2_db}/pdb_mmcif/mmcif_files \
+        --obsolete_pdbs_path=${params.alphafold2_db}/pdb_mmcif/obsolete.dat \
         --random_seed=53343 \
+        --use_gpu_relax \
         $args
 
     cp "${fasta.baseName}"/ranked_0.pdb ./"${fasta.baseName}".alphafold.pdb
@@ -71,6 +78,7 @@ process RUN_ALPHAFOLD2 {
     echo -e Positions"\\t"rank_0"\\t"rank_1"\\t"rank_2"\\t"rank_3"\\t"rank_4 > header.tsv
     cat header.tsv plddt.tsv > ../"${fasta.baseName}"_plddt_mqc.tsv
     cd ..
+    rm -rf "\${RUNTIME_TMP}"
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

diff --git a/modules/local/run_alphafold2_msa.nf b/modules/local/run_alphafold2_msa.nf
@@ -4,11 +4,13 @@
 process RUN_ALPHAFOLD2_MSA {
     tag   "$meta.id"
     label 'process_medium'
-    debug true
 
-    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'docker://nfcore/proteinfold_alphafold2_msa:1.0.0' :
-        'nfcore/proteinfold_alphafold2_msa:1.0.0' }"
+    // Exit if running this module with -profile conda / -profile mamba
+    if (workflow.profile.tokenize(',').intersect(['conda', 'mamba']).size() >= 1) {
+        error("Local RUN_ALPHAFOLD2_MSA module does not support Conda. Please use Docker / Singularity / Podman instead.")
+    }
+
+    container "nf-core/proteinfold_alphafold2_msa:1.1.1"
 
     input:
     tuple val(meta), path(fasta)
@@ -20,7 +22,7 @@ process RUN_ALPHAFOLD2_MSA {
     path ('mgnify/*')
     path ('pdb70/*')
     path ('pdb_mmcif/*')
-    path ('uniclust30/*')
+    path ('uniref30/*')
     path ('uniref90/*')
     path ('pdb_seqres/*')
     path ('uniprot/*')
@@ -35,33 +37,34 @@ process RUN_ALPHAFOLD2_MSA {
 
     script:
     def args = task.ext.args ?: ''
-    def db_preset = db_preset ? "full_dbs --bfd_database_path=${params.bfd_dir_path}${params.bfd_metaclust_clu_complete_id30_c90_final_seq_sorted_opt_name} --uniclust30_database_path=${params.uniclust30_dir_path}${params.uniclust30_db_name}" :
-        "reduced_dbs --small_bfd_database_path=${params.small_bfd_path}${params.bfd_first_non_consensus_sequences_name}"
+    def db_preset = db_preset ? "full_dbs --bfd_database_path=${params.alphafold2_db}/bfd/bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt --uniref30_database_path=${params.alphafold2_db}/uniref30/UniRef30_2021_03" :
+        "reduced_dbs --small_bfd_database_path=${params.alphafold2_db}/small_bfd/bfd-first_non_consensus_sequences.fasta"
     if (alphafold2_model_preset == 'multimer') {
-        alphafold2_model_preset += " --pdb_seqres_database_path=${params.pdb_seqres_dir_path}${params.pdb_seqres_txt_name} --uniprot_database_path=${params.uniprot_dir_path}/${params.uniprot_fasta_name} "
+        alphafold2_model_preset += " --pdb_seqres_database_path=${params.alphafold2_db}/pdb_seqres/pdb_seqres.txt --uniprot_database_path=${params.alphafold2_db}/uniprot/uniprot.fasta "
     }
     else {
-        alphafold2_model_preset += " --pdb70_database_path=${params.pdb70_dir_path}${params.pdb70_name} "
+        alphafold2_model_preset += " --pdb70_database_path=${params.alphafold2_db}/pdb70/pdb70_from_mmcif_200916/pdb70 "
     }
     """
-    #if [ -f ${params.pdb_seqres_dir_path}/${params.pdb_seqres_txt_name} ]
-    #    \$PDB_SEQRES_TEMP=\$(mktemp --directory)
-    #    cp ${params.pdb_seqres_dir_path}${params.pdb_seqres_txt_name} \${PDB_SEQRES_TEMP}/
-    #    then sed -i "/^\\w*0/d" \$PDB_SEQERS_TEMP/${params.pdb_seqres_txt_name}
-    #fi
+    RUNTIME_TMP=\$(mktemp -d)
+    if [ -f ${params.alphafold2_db}/pdb_seqres/pdb_seqres.txt ]
+        cp ${params.alphafold2_db}/pdb_seqres/pdb_seqres.txt \${RUNTIME_TMP}
+        then sed -i "/^\\w*0/d" \${RUNTIME_TMP}/pdb_seqres.txt
+    fi
     python3 /app/alphafold/run_msa.py \
         --fasta_paths=${fasta} \
         --model_preset=${alphafold2_model_preset} \
         --db_preset=${db_preset} \
         --output_dir=\$PWD \
         --data_dir=\$PWD \
-        --uniref90_database_path=${params.uniref90_dir_path}/${params.uniref90_fasta_name} \
-        --mgnify_database_path=${params.mgnify_database_path}/${params.mgy_clusters_fasta_name} \
-        --template_mmcif_dir=${params.template_mmcif_dir} \
-        --obsolete_pdbs_path=${params.obsolete_pdbs_path}  \
+        --uniref90_database_path=${params.alphafold2_db}/uniref90/uniref90.fasta \
+        --mgnify_database_path=${params.alphafold2_db}/mgnify/mgy_clusters_2022_05.fa \
+        --template_mmcif_dir=${params.alphafold2_db}/pdb_mmcif/mmcif_files \
+        --obsolete_pdbs_path=${params.alphafold2_db}/pdb_mmcif/obsolete.dat  \
         $args
 
     cp "${fasta.baseName}"/features.pkl ./"${fasta.baseName}".features.pkl
+    rm -rf "\${RUNTIME_TMP}"
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

diff --git a/modules/local/run_alphafold2_pred.nf b/modules/local/run_alphafold2_pred.nf
@@ -3,12 +3,16 @@
  */
 process RUN_ALPHAFOLD2_PRED {
     tag   "$meta.id"
-    label 'process_medium', 'gpu_compute'
+    label 'process_medium'
+    label 'gpu_compute'
+
+    // Exit if running this module with -profile conda / -profile mamba
+    if (workflow.profile.tokenize(',').intersect(['conda', 'mamba']).size() >= 1) {
+        error("Local RUN_ALPHAFOLD2_PRED module does not support Conda. Please use Docker / Singularity / Podman instead.")
+    }
+
+    container "nf-core/proteinfold_alphafold2_split:1.1.1"
 
-    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'docker://nfcore/proteinfold_alphafold2_split:1.0.0' :
-        'nfcore/proteinfold_alphafold2_split:1.0.0' }"
-    echo 'true'
     input:
     tuple val(meta), path(fasta)
     val   db_preset
@@ -19,7 +23,7 @@ process RUN_ALPHAFOLD2_PRED {
     path ('mgnify/*')
     path ('pdb70/*')
     path ('pdb_mmcif/*')
-    path ('uniclust30/*')
+    path ('uniref30/*')
     path ('uniref90/*')
     path ('pdb_seqres/*')
     path ('uniprot/*')
@@ -36,18 +40,15 @@ process RUN_ALPHAFOLD2_PRED {
     script:
     def args = task.ext.args ?: ''
     """
-    echo \$PWD
-    #if [ -d ${params.alphafold2_params_path} ]; 
-    #then
-    ln -r -f -s ${params.alphafold2_params_path}* params/
-    #fi
+    if [ -d ${params.alphafold2_db}/params/ ]; then ln -r -s ${params.alphafold2_db}/params params; fi
     python3 /app/alphafold/run_predict.py \
         --fasta_paths=${fasta} \
         --model_preset=${alphafold2_model_preset} \
         --output_dir=\$PWD \
         --data_dir=\$PWD \
         --random_seed=53343 \
         --msa_path=${msa} \
+        --use_gpu_relax \
         $args
 
     cp "${fasta.baseName}"/ranked_0.pdb ./"${fasta.baseName}".alphafold.pdb